AI-enhanced knowledge ecosystems: a paradox framework for understanding knowledge sharing and hiding in generative AI

Organisations increasingly rely on artificial intelligence to capture, synthesise and generate knowledge, yet the governance consequences of this shift remain poorly theorised. This paper develops a conceptual framework for AI-enhanced knowledge ecosystems, arguing that AI does not simply augment knowledge management practice but reshapes what counts as organisational knowledge. Drawing selectively on work in knowledge management, AI and paradox theory, the framework sets out five dimensions of such ecosystems and four paradoxes: sharing and hiding, transparency and opacity, control and autonomy, exploitation and exploration. These tensions are treated as generative forces to be navigated, not problems to be solved, since managing them adaptively can support learning, innovation and resilience. Three scenarios extend the analysis to 2050 under different governance conditions. The paper contributes testable propositions on AI-specific knowledge mechanisms and an agenda for empirical validation at ecosystem level, with implications for knowledge governance and responsible AI deployment.